We hypothesize that 1) an important cause of severe asthma is altered inflammatory responses that are partially related to sequence variants in genes that regulate bronchial inflammation, pulmonary function or affect structural components in the airways and 2) a subset of patients develops severe asthma because of pharmacogenetic responses to pharmacologic agents. Our first aim is to understand the longitudinal characteristics that are important in the development of severe asthma using both standard and cluster approaches. For each of the first 3 years, we will recruit and characterize 96 subjects (60 % with severe asthma, 25% children; assuming 20% loss to follow up). Baseline and 3 year studies will include PFTs, maximum bronchodilator reversibility, bronchial responsiveness to methacholine, comprehensive questionnaires, blood for eosinophils, neutrophils, DNA and total serum IgE levels, induced sputum, eNO and CT imaging. In a substudy, investigative bronchoscopy will be performed at baseline and year 3. An evoked phenotype will be assessed two weeks after administration of triamcinolne acetonide injectable suspension by evaluating changes in baseline phenotypes including lung function, bronchial responsiveness, induced sputum and biomarkers. Subjects will be reassessed yearly (spirometry and reversibility, eNO, sputum induction and questionnaires) with additional telephone contact every six months. Each subject will be asked to return for an additional visit when they are experiencing an exacerbation. All studies with the exception of bronchoscopy and CT imaging will be performed in children. Our second aim is to determine genetic and biomarker predictors of baseline phenotypes and their change over time. New subjects will be assigned to a current SARP asthma cluster. When the total population has been studied, a new cluster analysis will be performed for comparison with the current clusters, including additional biomarkers and CT imaging. Individual changes in cluster assignment will be assessed longitudinally. Biomarker and genetic analysis (including the role of rare variants) will be performed using the clusters and longitudinal data including lung function, as well as pharmacogenetic analysis of the evoked systemic steroid phenotype.